FILTERAgents & automation
JEVBOT Trading Desk Audit and Rebuild
JEVBOT@JevBotApp
JEVBOT has audited its initial trades, identified common failure patterns, and rebuilt its trading filters and mechanisms to improve performance and risk management.
TakeA coin must clear hard rules BEFORE Jev is even asked: minimum age, real wallets, human-sized trades, volume that matches its market cap, buyers in control, no vertical candles. 👥 Holder data.
🤖 JEVBOT desk update: we audited every trade and rebuilt the filters
JEVBOT has been live for a few hours: real wallet, real fees, real trades, all public. The first 45 closed trades were not good: 33% win rate. So we did what a desk should do: we opened the book and studied every loser.
What the data showed:
• Coins under 15 minutes old were 20 of 45 trades and most of the damage (bundled launches that dump within a minute)
• "Fake pools": $1M+ market caps with ~$300 of volume, made of $3 bot trades
• Coins traded by only a handful of wallets (one went to -98% when liquidity was pulled)
• Vertical 5-minute candles = buying the top
• Jev was judging rug risk without seeing who holds the coin
What changed:
🔎 The Screen. A coin must clear hard rules BEFORE Jev is even asked: minimum age, real wallets, human-sized trades, volume that matches its market cap, buyers in control, no vertical candles.
👥 Holder data. Jev now sees holder count, top-10 concentration, and dev / bundler / sniper bags. Coins that are top-heavy or bundled never reach it.
💱 Round-trip check. Before every buy the desk quotes the buy AND the sell back. Thin or taxed liquidity = no trade.
⚡ 15-second watcher. Positions are re-marked every 15s (not every 60s): stop-loss, hard stop and a trailing stop that locks gains once a trade has run.
🔥 Buyback & burn. Every few scans Jev decides whether the desk buys back $JEVBOT and burns it. 6 buybacks so far, 3.18M tokens burned, receipts on chain.
Backtested on our own 45 trades, the new screen would have skipped almost all of the big losers and kept the best winners. Small sample, no promises. The desk will trade less often and be pickier. Memecoins are memecoins, and it can still lose.
Same rule as day one: nothing is labelled Jev that Jev didn't answer, and every rule the desk follows is printed on the site.
Watch it think, live: jevbot.app
Related builds
